|
|
|
|
const puppeteer = require('puppeteer');
|
|
|
|
|
const { pathOr, maxBy } = require('ramda');
|
|
|
|
|
const path = require('path');
|
|
|
|
|
const fs = require('fs').promises;
|
|
|
|
|
|
|
|
|
|
const { graphql_endpoint } = require('./constants');
|
|
|
|
|
const {
|
|
|
|
|
create_images_directory,
|
|
|
|
|
parse_args,
|
|
|
|
|
read_previous_events,
|
|
|
|
|
} = require('./logic');
|
|
|
|
|
|
|
|
|
|
const { page_ids, output, events: event_file } = parse_args(
|
|
|
|
|
process.argv.slice(2),
|
|
|
|
|
);
|
|
|
|
|
|
|
|
|
|
(async () => {
|
|
|
|
|
create_images_directory('./img');
|
|
|
|
|
|
|
|
|
|
const previous_events = await read_previous_events(event_file);
|
|
|
|
|
|
|
|
|
|
const browser = await open_browser();
|
|
|
|
|
|
|
|
|
|
let events = [];
|
|
|
|
|
|
|
|
|
|
for (let page_id of page_ids) {
|
|
|
|
|
let scraping = true;
|
|
|
|
|
const facebook_page = await browser.newPage();
|
|
|
|
|
const upcoming_events = register_upcoming_events_listener(
|
|
|
|
|
graphql_endpoint,
|
|
|
|
|
facebook_page,
|
|
|
|
|
)
|
|
|
|
|
.then((upcoming_events) => {
|
|
|
|
|
scraping = false;
|
|
|
|
|
return upcoming_events;
|
|
|
|
|
})
|
|
|
|
|
.catch((err) => {
|
|
|
|
|
console.error(err);
|
|
|
|
|
scraping = false;
|
|
|
|
|
});
|
|
|
|
|
|
|
|
|
|
await facebook_page.goto(page_id);
|
|
|
|
|
|
|
|
|
|
while (scraping) {
|
|
|
|
|
await facebook_page.evaluate(() =>
|
|
|
|
|
window.scrollBy(0, window.innerHeight),
|
|
|
|
|
);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
const responses = await upcoming_events;
|
|
|
|
|
const nodes = responses.reduce(
|
|
|
|
|
(res, current) => [...res, ...current.edges],
|
|
|
|
|
[],
|
|
|
|
|
);
|
|
|
|
|
const new_events = nodes.map(map_event);
|
|
|
|
|
|
|
|
|
|
events = merge_edges(new_events, events);
|
|
|
|
|
events = events.filter(
|
|
|
|
|
({ event_id }) =>
|
|
|
|
|
previous_events.find(
|
|
|
|
|
(previous_event) => event_id === previous_event.event_id,
|
|
|
|
|
) === undefined,
|
|
|
|
|
);
|
|
|
|
|
|
|
|
|
|
events = await Promise.all(
|
|
|
|
|
events.map(async (event) => {
|
|
|
|
|
const event_page = await browser.newPage();
|
|
|
|
|
const event_data = await load_event(event_page, event.event_id);
|
|
|
|
|
return {
|
|
|
|
|
...event_data,
|
|
|
|
|
...event,
|
|
|
|
|
};
|
|
|
|
|
}),
|
|
|
|
|
);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
events = await Promise.all(
|
|
|
|
|
events.map(async (event) => {
|
|
|
|
|
const images = await save_images(event);
|
|
|
|
|
delete event.image;
|
|
|
|
|
return {
|
|
|
|
|
images,
|
|
|
|
|
...event,
|
|
|
|
|
};
|
|
|
|
|
}),
|
|
|
|
|
);
|
|
|
|
|
|
|
|
|
|
if (output === null) {
|
|
|
|
|
const all_events = [...events, ...previous_events]
|
|
|
|
|
.map((event) => {
|
|
|
|
|
const start = pathOr(null, ['date', 'start'], event);
|
|
|
|
|
if (start !== null) {
|
|
|
|
|
try {
|
|
|
|
|
event.date.start = new Date(start);
|
|
|
|
|
} catch (e) {
|
|
|
|
|
console.error(e);
|
|
|
|
|
return event;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return event;
|
|
|
|
|
})
|
|
|
|
|
.sort((a, b) => {
|
|
|
|
|
const b_date = b.date.start;
|
|
|
|
|
const a_date = a.date.start;
|
|
|
|
|
if (a_date > b_date) {
|
|
|
|
|
return 1;
|
|
|
|
|
}
|
|
|
|
|
if (a_date == b_date) {
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
return -1;
|
|
|
|
|
});
|
|
|
|
|
|
|
|
|
|
console.log(JSON.stringify(all_events));
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
process.exit();
|
|
|
|
|
})();
|